Calibration lights for integrating sphere

I’m preparing to build my own integration sphere (styrofoam ball 50 cm diameter) and I need a calibration point. Instead of buying expensive brand lights with more or less exact lumen info I intend to build my own cheap calibration light(s) and have them measured by other calibrated spheres later. Have some questions though:

1.) Do I need more than one calibration light - for instance with different reflector diameters, tints and output levels? I guess I need 2 tints since I won’t buy a professional luxmeter, and cheaper ones are said to react different on different tints. I also want to measure lights with high output (10000 lumens or more). Do I need a high output calibration light then?

2.) What other requirements are there? E. g., is it important to have flat bezels or black bezels (no stainless steel or bare aluminium).

3.) I intend to use a 105C or so with 6x or 4x 7135, in order to always maintain the same output with one cell. Is this stable?

Thanks in advance.

Hi Mike,
I have some experience in making them.

First, indeed some high end flashlights (Nitecore and Sunwayman comes to mind), on the medium or low levels (not the highest level!) , have some brilliantly flat regulated output. Selfbuilt’s tests at CPF are a good reference for that. My standard calibration light is a SWM D40A at high 550lm setting (turbo is not stable, probably direct drive), which uses 4xAA NiMh cells with the extra advantage that because of the higher voltage overhead compared to li-ion it stays in regulation for most of the drainage of the cells. The disadvantage of the D40A is its wide head with SS bezel, it does not fit in the smallest hole of my integrating sphere, and it interferes more with the reflectivity (multiplier) of the sphere than a small head light with black bezel (in my sphere I can measure it and correct for that using an inbuild permanent calibration light).

Making your own calibration light is a very good possibility by using a host (tube style 18650 light, black bezel is most practical I think) with proper heatsinking and using a driver with just one or two 7135 chips. I have never found any error doing the calibration with a low output (say 100-200lm) calibration light while measuring high output lights (luxmeters are pretty lineair, even the cheap ones). And making a high output calibration light brings all sorts of extra problems and less constant output. Btw, I found that not all 7135 chips are of same quality, some have an output current that is quite sensitive for heat.

If you first build the sphere, and then the calibration light, it is dead easy to check how well the light works for its use. Make a runtime/output curve for your calibration light and decide which part of the curve is flat enough for use. (the first 20 seconds are usually not flat enough).

Thanks a lot for the detailed information, djozz. I already thought about using black Convoy S2+ for this purpose, with a reduced number of 7135. I will also check my Thrunite TN4A which seems to be similar to your D40A.

If I remember right it was you who tested cheap luxmeters and found out that with cool tints the readings were somewhat higher. So wouldn’t it be a good idea to have at least two calibration lights, e.g. one with a 1A tint and another one with a warmer tint, which could be used as reference for measurement of lights with similer tints?

The measurement difference between a cool white and neutral white flashlight was just 2% if I remember it well, a bit more for warm white. If you want an accuracy for your sphere that is in that region, there is more to take care of than just the cool white/neutral white error. As an example you may want an internal reference light inside your sphere to correct for reflectivity errors caused by the varying flashlights in the entrance hole (I described building a very good working reference light in my sphere#4 thread, using a 9V block battery, a 5mm led, a resistor and a LM317 voltage/current regulator).

Besides, how are you going to find out what the real output is for your two calibration lights, to be able to see the error in your own set-up?

Ok, my memory was wrong, I thought the difference was higher. No, I don’t need that accuracy for my sphere, I just want a more convenient method of measuring my modded lights than using our bathroom as a “integration cube” as I do currently. My family would appreciate this …
To measure my calibration lights I probably will ask one or two german TLF members with spheres for help, once the lights are done and stable.

Flashy Mike, how did (or is) you sphere calibration lights work out? I posted similar questions here, and I know I remember reading responses to my answers but I can’t find them right now (perhaps in another thread?). I remember being told to go with an XP-G2 due to it’s higher vF which I have done. I have a Convoy M1 XP-G2 with a 6x7135 Nanjg driver custom flashed with STAR firmware to modes of 5–33–100% for this purpose (I generally don’t hold “high” mode). I then did a “recalibration of my sphere” borrowing the original Fenix HL-55 headlamp I used again and this time took measurements of many lights with nanjg “current regulated” drivers being sure to use fully charged cells.

I’ve had quite a few measurement sessions now since then and for now in addition to using my “sphere calibration check light” (i.e. the M1 XP-G2) I also measure at least some of those same nanjg driver lights. What I am seeing in actual use is that my XM-L2 lights (Convoy M1, 3.04A driver 8x7135 custom flashed to 2-modes, 35–100, and my mule headlamp with stock 1.4A 3-mode nanjg driver) are steadier and more consistent than the XP-G2. Even the M1 XM-L2 @ 3.04A is surprisingly steady. Overall I’m seeing some decent consistency between calculated lumens on these same lights from session to session.

Just curious how yours turned out.

-Garry

I’m looking to build one also Mike can u post a link on where to buy the sphere ? Or is building a PVC type lightbox better ?

I think Mike just did what I did - bought two half spheres and built his own. I bought mine from a local JoAnn Fabrics store, but here it is online: http://www.joann.com/smoothfoam-12-inch-half-ball/12384335.html (12" - 30.5cm). Here are pics of mine: https://budgetlightforum.com/t/-/33478/64 posted in djozz's build thread (do read through that whole thread). I personally would avoid the PVC tube method unless you follow "what Texas_Ace did":https://budgetlightforum.com/t/-/43380. My 3" PVC Tube build gave wildly inconsistent results and was far more difficult to build.

Pic of mine finished:

Pic of mine in use:

Problem I have: the velcro doesn't hold the luxmeter extremely tight so velcro lets the luxmeter back off over time (I press the velcro for a tight mount each session) and this was the important reason for my "calibration check light".

-Garry

I have an old ikea cheap globe lamp that has a plastic ball. Approx 12 inches tall. It has a hole top and bottom. I think it will make a good integrating sphere. How does the light leakage affect the readings on the styrofoam spheres?

Isn’t the idea to keep to keep the light in? I can tell you that ambient room lighting coming from the outside hardly has any affect on mine. If I turn my luxmeter on while in the dark, then flip the room lights on I hardly see a change on the meter (maybe from 0.0 to 0.2 or 0.3). I’m sure this would depend on how bright your room lights are though.

-Garry

Actually I haven’t built my sphere yet. Too busy with other projects currently …

Yes Gary, the idea is to keep the light in. The reason I ask is due to the light emitting from the ball in the pictures above, light leakage so I’m wondering how that affects the readings.

Something similar to this is what I have and intend to transform at some stage
http://www.aboveandbeyond.co.uk/Products/SKU6308/the-traveller-globe?utm_source=froogle&utm_medium=feed&utm_campaign=froogle-campaign&gclid=COPZ5fnFldICFW6x7QodPW4JkA

The “map” peels off easily leaving a diffuse plastic sphere beneath

Flashy Mike: Hmm . . . I see. I’ve been wondering about mounting an XM-L2 to a large PC heatsink and driving it with an adjustable power supply like an LM2596 board or even one of these adjustable boards with digital display I have on-hand. I wonder if I would get better consistency with it. I like the idea of not relying on battery power.

-Garry

I experimented with mine taking readings (a year ago) as is and then covering the sphere completely with a thick black towel and the difference was negligible. Looking back at my notes, testing with a Convoy S2 with XM-L2 and stock 3 mode Nanjg 2.1A I saw lux readings of (@30secs):

Dark room, no towel:
Low 58.0
Med 478
High 1090

Sphere placed directly under a 20watt CFL (75watt equiv.) light bulb (directly overhead of sphere opening):
Low 57.5
Med 472
High 1085

Dark Room, Sphere wrapped in towel:
Low 58.9
Med 484
High 1104

Keep in mind that I did my calibration “without towel” so I would think adding the towel/covering wouldn’t be a good idea.

-Garry

I did a quick test of applying various sphere multipliers to readings I took on a specific date to see how much spread there would be in my readings. I use my “calibration check lights” every session, so I’m actually computing new multipliers each session (remember, I can’t count on my luxmeter being in the same perfect alignment each time, so no perfectly constant “multiplier” to use). In real world use (over the past month now) my multipliers have only ranged from 0.381 to 0.387 with 0.3865 being what seems the most consistent. Even across my measurements last year I only saw a range of 0.38 to 0.405. So here’s what I calc’d up:

Note that my “Expected Lumens” values are not necessarily known with 100% certainty. And of course the “Lumens Multiplier #3” column will closely match those values because that’s a multiplier very close to the one used to determine those values (so don’t let that fool you). This was merely done to see how far off I would be if I determined a “bad” multiplier.

-Garry

Thanks for posting that Garry, it is good to get a grip on the actual variation that the measurments have.

About keeping the light in: for good integration you need the light to bounce inside the sphere multiple times so a high reflection internally is a good thing. On the other hand, if you are over a certain minimum amount of reflectivity that is needed to get the integration good, it is no problem at all to loose the rest of the light, through the wall or through the various entrance and exit holes, the real concern is that this light loss remains fixed so that your multiplier will not alter.

Hey djozz, any comment on what I said in” post #13 above”:Calibration lights for integrating sphere - #13 by garrybunk? Better option?

-Garry

My intuition is that I it will work well. The problem is that I have no real knowledge of electronics/power supplies, so the only way for me to find out if that way produces more constant light sources is to actually build it and test it. Which is just as well possible for anyone else with a sphere.

It is just like my trick with the 9V battery, a resistor, an LM317 and a 5mm led: I had no idea if it would work but it tested as an extremely constant light source (although with a very low output of only a few lumen, which would not neccessarily work for everyone).

Thanks djozz. Perhaps one of these days I’ll try it out. I do already have an XM-L2 mounted to a large heatsink, but it’s not on a copper dtp board and it’s a setup I use to test firmware flashing. I should also be careful to use a known bin so that way I could test against the Cree datasheets and tests by others.

-Garry